Clean Up Grainy Video
Reduce the sensor noise in low-light footage with FFmpeg hqdn3d, at a strength you choose, without sending the file anywhere.
Drag & drop your video here, or click to browse
Max file size: ~2 GB (memory permitting)
How to Use — Clean Up Grainy Video
Load the noisy clip
Drag in an MP4, MOV or WebM. The file stays on your device — nothing is uploaded.
Pick a strength
Light, Medium or Strong map to hqdn3d=2:1.5:3:2.25, 4:3:6:4.5 and 8:6:12:9. The tool shows the estimated run time for your resolution and duration before you press anything.
Denoise, then compare
Run it and pause the result on a detailed shot — hair, brickwork, fabric. If the texture has gone waxy, the setting is too strong for this footage.
Compress after, not before
Download the MP4 and compress it next. Noise is random, so an encoder has to spend bits on it every frame; removing it first buys real bitrate for the picture.
Popular task presets
Best for / not for
Best for
- Low-light phone and camera footage where sensor noise crawls across flat areas like walls, sky and skin.
- Grainy clips heading for a compressor or an upload — removing the noise first buys real bitrate for the picture.
- Old camcorder, webcam and screen-capture video with visible speckle.
Not for
- Recovering detail. hqdn3d averages pixels together; it can only remove information, never add it.
- Compression artefacts already baked in — blocking and banding are not random, so temporal averaging just blurs them.
- Fast action at Strong: the doubled temporal values leave trails behind anything that moves quickly.
When to use the Video Denoiser
- A phone clip shot indoors at night, where the shadows fizz with sensor noise.
- Interview or vlog footage lit with one lamp — Light or Medium takes the crawl off faces without turning them waxy.
- Grainy stock or archive footage that has to survive a second encode before publishing.
- Anything about to be compressed: denoise first, compress second, and the same bitrate buys a cleaner picture.
Supported formats & limits
| Input containers | MP4, MOV, WebM, MKV |
|---|---|
| Input codecs | H.264, H.265/HEVC, VP8, VP9, MPEG-4, MJPEG. Not AV1 — this build has no software AV1 decoder. |
| Output container | MP4 — plays on iOS, Android, YouTube, Instagram and X |
| Output codec | H.264 (libx264, veryfast, CRF 23 default quality) — a full re-encode, so the file size changes. Audio is copied from MP4/MOV sources and re-encoded to AAC 192 kbps from anything else. |
| Filter | hqdn3d at three strengths: Light 2:1.5:3:2.25, Medium 4:3:6:4.5 (the filter default), Strong 8:6:12:9 |
| Speed | Measured in this build: 2.9 s for 3 seconds of 720p on a desktop. Time scales with pixel count, so 1080p is roughly 2.3× that; phones are several times slower. |
| Max file size | Up to ~2 GB (limited by browser memory) |
| Cost | Free for any use. No signup. No watermark. |
Video Denoiser vs. the usual alternatives
| Feature | This tool | VEED (free) | Kapwing (free) | CapCut Online |
|---|---|---|---|---|
| Processing model | Runs locally in your browser | Upload-based project editor | Upload-based project editor | Upload-based online editor |
| File limits | No upload cap; practical limit is browser memory | Plan-specific upload limits | Plan-specific upload and export limits | Feature- and account-specific limits |
| Watermark on output | No watermark added | Free exports include a VEED watermark | Free exports include a Kapwing watermark | Standard edits can be watermark-free; templates/assets may add branding |
| Signup / account | No account for tools | Workspace/account flow | Workspace/account flow | CapCut account flow |
| Works offline | Yes after cache, subject to browser support | No | No | No |
| Best for | Private one-step file operations | Full editor, templates, AI tools | Collaboration, templates, AI tools | Social templates and timeline editing |
Vendor plan limits were checked on April 29, 2026 and can change by region, account state, and export option. Verify critical limits on the vendor pricing/help page before relying on them.
Why use this video denoiser
- The three strengths are the real hqdn3d numbers, printed on the buttons, so you can reproduce the run in a terminal later.
- It estimates the run time for your resolution and duration before you start, from a measurement taken in this same engine.
- It says what denoising costs instead of promising a repair: every setting removes grain and fine texture together.
- 100% private — the file is denoised in your browser through WebAssembly and never uploaded.
Task-focused FAQ
Will denoising sharpen a blurry video?
No. hqdn3d only averages pixels together, which softens; nothing here adds detail back.
Which strength should I start with?
Light. Compare the result against the original on a detailed shot, and only move up if noise is still the first thing you notice.
Does it work on compression artefacts?
Not really. Blocking and banding are not random, so the temporal averaging that kills sensor noise just blurs them.
Frequently Asked Questions
How does the denoiser work?
FFmpeg's hqdn3d averages each pixel with its neighbours inside the frame and with the same pixel in nearby frames. Noise is random and cancels out under that averaging; real detail is not, so some of it survives. It all runs locally through WebAssembly.
What do Light, Medium and Strong change?
The four hqdn3d parameters: luma spatial, chroma spatial, luma temporal, chroma temporal. Light 2:1.5:3:2.25, Medium 4:3:6:4.5 (the filter default), Strong 8:6:12:9. Each step doubles the averaging — Strong also doubles the temporal values, so fast movement can leave trails.
Will I lose detail?
Yes, at every setting. A denoiser cannot tell grain from fine texture, so it takes some of both. Start at Light and only go up if noise is still the first thing you notice.
Denoise before or after compressing?
Before. An encoder cannot predict random noise from the previous frame, so it spends bits describing it on every frame. Strip it first and that bitrate goes to the picture — often a smaller file that also looks better. Denoising after compression only blurs artefacts that are already baked in.
How long will it take?
Every frame is re-encoded with libx264 veryfast at its default quality (CRF 23). Measured here: 2.9 s for 3 s of 720p on a desktop — roughly real time at 720p, about 2.3× that at 1080p (the pixel count), several times slower on a phone. The strength barely changes the time. The tool estimates it for your file before you start.
What comes out?
An MP4 with H.264 video. The video is fully re-encoded with libx264 at its default quality (CRF 23, veryfast) — denoising has to rewrite the pixels — so expect a fresh lossy encode and a different file size. Audio from an MP4 or MOV source is copied bit-for-bit; audio from a WebM or MKV is re-encoded to AAC 192 kbps, because Opus, Vorbis and FLAC are not legal inside an MP4.